Application of geostationary satellite data to derive normalized difference vegetation index NDVI

1
Application of geostationary satellite data to
derive normalized difference vegetation index
(NDVI)

• Peter Romanov, CICS-University of Maryland
• Hui Xu, IMSG Inc.

2
Getting ready for GOES-R

• GOES-R ABI first GOES instrument capable of
  monitoring NDVI
• Objective Evaluate potentials for NDVI mapping,
  demonstrate benefits, identify problems, develop
  techniques.
• MSG (Meteosat-8-9) SEVIRI GOES-R ABI prototype

GOES Geostationary Operational Environmental
Satellite ABI Advanced Baseline Imager MSG
Meteosat Second Generation SEVIRI Spinning
Enhanced Visible and Infrared Imager
3
Normalized Difference Vegetation Index
NDVI ( Rnir - Rvis ) / ( Rnir Rvis )
Rnir Near IR reflectance (0.9 µm) Rvis
Visible reflectance (0.6 µm)

• Larger NDVI means greener vegetation
• Clouds and snow Low positive/negative NDVI
• NDVI related to leaf area index and biomass

University of Arizona
4
NDVI monitoring at NOAA

• Afternoon NOAA AVHRR data
• One observation per day
• Maximum NDVI compositing
• Weekly maps since 1981
• 4, 8 and 16 km resolution
• Global coverage

5
AVHRR NDVI Problems

• Spurious short-term variability due to
• - Remaining clouds
• - Reflectance anisotropy variable
  observation geometry
• Spurious trends due to
• - Satellite orbital drift
• - Inaccurate sensor calibration

Sudan, year 2007, GVI-x 16 km
6
Geo satellites expected benefits for NDVI

• Frequent repeat cycle
• - More clear sky views
• - Better daily spatial coverage
• - Better temporal resolution
• - Better cloud identification
• Fixed satellite position
• - Less variation in the measurement geometry
• - Less spurious variability in NDVI time
  series

7
MSG SEVIRI and GOES-R ABI
GOES-R ABI
MSG SEVIRI
15 min scans Spatial resolution - 0.5 km
visible - 2 km all other Position 75W and
135W Launch 2014
15 min scans Spatial resolution - 1 km HRV
(visible) - 4 km all other Position 0E
Operational since 2005
8
MSG data collection and processing
Images are collected every 30 minutes Europe and
Africa, since March 2006 Full disk, since
February 2007 NDVI estimates - Daily and
weekly max NDVI compositing - Cloud mask has
been developed
False color RGB full disk image
9
Effect of daily compositing
Image time 13.15 UTC
June 23, 2007
10
Weekly compositing
One image per day (13.15 UTC) Max NDVI compositing
All daily images Max NDVI compositing
Week June 16-23, 2007
11
Geometry of observations and NDVI

• Satellite zenith angle is specific for every
  location and does not change with time
• Most observations are performed in the
  backscatter

June 23, 2007
Reflectance and NDVI
Solar elevation and relative azimuth
12
Diurnal change of reflectance and NDVI
July 19, 2006
13
MSG Time of Max NDVI
Jul 18-24, 2006
Oct 27 Nov 2, 2006
Jan 10-16, 2007
Optimal conditions for compositing Time 8h
to 13h local time Solar elevation above 200 -250
14
NDVI NOAA AVHRR vs MSG SEVIRI
Weekly max NDVI composite averaged within 1x1 deg
area
MSG SEVIRI Derived from half-hourly images at 4
km resolution
NOAA-18 AVHRR (derived from 16 km GVIx)
July 16-22 2006
15
MSG NDVI Satellite zenith angle effect
MSG and AVHRR NDVI vs MSG satellite zenith angle
16
AVHRR vs MSG NDVI time series
17
AVHRR vs MSG NDVI time series
Year 2007
18
AVHRR vs MSG NDVI time series
19
NDVI daily change
Year 2007
Ch.1 (visible) reflectance
Ch.2 (near IR) reflectance
Lat 9.4 S Lon 25.5 E
Diurnal variation of ch.1, ch.2 reflectance and
NDVI over savanna during 8-day period as observed
from MSG.
NDVI
20
Summary

• Geostationary satellites can improve vegetation
  monitoring
• Better temporal resolution, hence potential for
  earlier identification of droughts
• Better consistency in time series hence better
  climatology
• Coverage is limited
• Estimates at solar and/or satellite zenith angle
  above 700 are hard to use
• Correction is required for satellite zenith
  angle
• Further improvement in retrievals requires
  proper bidirectional correction
• Still some variations will occur due to
  atmospheric (aerosol) effects